View this email in your browser
2019-04-01 | J++ Nyhetsbrev #72
A study on criminal justice teaches us how to do data driven journalism on court cases
This week's newsletter focuses on a new study from Dr. Katrin Lainpelto at Stockholm University, which examines how criminal courts treat suspects based on if they have a Swedish background (children of parents born in Sweden) or an immigrant background (children of parents born outside of Sweden) in the context of sexual harassment.

We will partly summarise the results of the study but also give our thoughts on what this study teaches us, when we want to do data driven journalism on court cases. 

But first some background.

In August last year, just weeks ahead of the Swedish general elections, the investigative TV show Uppdrag granskning reported that they had gathered statistics from court cases, showing that rape convicts were more likely to be foreign born than not. One of the explanations to this over-representation, was reported to be cultural differences, and more specifically differences in view of sexuality.

Last week, when Dr. Katrin Lainpelto presented her study, she put this question into a new light. Lainpelto, who cited the Uppdrag granskning report as a reason why this subject needs to be examined more closely, asked:
  • Is there a difference in how courts treat suspects based on if they have a Swedish background or an immigrant background, and if so, what are possible explanations to this difference?
To answer these questions, Latinpelto used an experimental survey and an analysis of the cited evidence in actual judgements and prosecutions.

In the survey, judges and lay judges were asked to answer questions regarding two fictional cases and interrogations of sexual harassment suspects, the only difference being the name of the suspects (Mudhafar/Erik) and the way they spoke Swedish (with an accent/without an accent).

Lainpelto's study can be read here (in Swedish), page 127-149:
Lainpelto's conclusions:
  • Young people with immigrant background (newly arrived people in particular) are convicted (and seemingly also prosecuted) on less and more questionable evidence when compared to young people with Swedish background.
  • Judges and lay judges experienced a greater feeling of discomfort towards Mudhafar than Erik.
  • Both judges and lay judges thought that Erik's version of events were more reliable than Mudhafar's, even though their stories were identical. They also thought that Erik’s story was more detailed.
  • In the end, 13 in 100 lay judges acquitted Mudhafar while 38 in 100 acquitted Erik, meaning that they were three times more likely to acquit Erik.
  • Lainpelto points at media discourse as one possible reason why foreigners and people with immigrant background are being treated differently by the Swedish justice system.
What we as journalists should learn from Lainpelto's study, if we want to do data driven journalism on court cases: 
  • Aggregating data on birth places, sentences and verdicts is seldom good enough. Court cases often need to be examined from a more qualitative perspective. What was the evidence that the court used to convict or acquit?
  • Disregarding what happened before a case reached the court is problematic. There may be systemic issues behind why someone is prosecuted or not prosecuted to begin with. 
  • Some crimes are more complex than others; the Swedish police “clear” about 7% of reported rape cases, compared to practically 100% of reported cases regarding drunken driving. We can’t use the same method to investigate both.
Do you want to replicate Rita Costa's story on gendered street names from last year? Giorgio Comai, who also is a part of the European Data Journalism Network, has made it easy for you by creating a library in R to automatically classify the gender of street names in any country. Giorgio has published the code and a notebook on how to use it.

The last couple of weeks have been dominated by more or less confusing Brexit-related votes in the British parliament. We think that the UK Institute for Government did a really good job in trying to make things a little bit more clear.

Follow Peter Krantz's exploratory code to extract all images (and possibly more in the future? :O) from a PDF-file here.
You have six days to submit your work to this year's round of the Data Journalism Awards. What are you waiting for? As always with the DJA, you can see submitted projects in real time.
J++ Stockholm |

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.
Copyright © *|CURRENT_YEAR|* *|LIST:COMPANY|*, All rights reserved.

Our mailing address is:

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.